Deficiency of lncRNA SNHG12 impairs ischemic limb neovascularization by altering an endothelial cell cycle pathway

SNHG12, a long noncoding RNA (lncRNA) dysregulated in atherosclerosis, is known to be a key regulator of vascular senescence in endothelial cells (ECs). However, its role in angiogenesis and peripheral artery disease has not been elucidated. Hind-limb ischemia studies using femoral artery ligation (FAL) in mice showed that SNHG12 expression falls readily in the acute phase of the response to limb ischemia in gastrocnemius muscle and recovers to normal when blood flow recovery is restored to ischemic muscle, indicating that it likely plays a role in the angiogenic response to ischemia. Gain- and loss-of-function studies demonstrated that SNHG12 regulated angiogenesis — SNHG12 deficiency reduced cell proliferation, migration, and endothelial sprouting, whereas overexpression promoted these angiogenic functions. We identified SNHG12 binding partners by proteomics that may contribute to its role in angiogenesis, including IGF-2 mRNA–binding protein 3 (IGF2BP3, also known as IMP3). RNA-Seq profiling of SNHG12-deficient ECs showed effects on angiogenesis pathways and identified a strong effect on cell cycle regulation, which may be modulated by IMP3. Knockdown of SNHG12 in mice undergoing FAL using injected gapmeRs) decreased angiogenesis, an effect that was more pronounced in a model of insulin-resistant db/db mice. RNA-Seq profiling of the EC and non-EC compartments in these mice revealed a likely role of SNHG12 knockdown on Wnt, Notch, and angiopoietin signaling pathways. Together, these findings indicate that SNHG12 plays an important role in the angiogenic EC response to ischemia.


Cell Culture and Transfection
Human umbilical vein endothelial cells (HUVECs; Lonza) were cultured in endothelial cell growth medium EGM-2 (Lonza). Cells that were utilized for experiments were passaged no more than six times. Peripheral blood mononuclear cells (PBMCs) were isolated by overlaying 1:2 whole blood/normal saline mixture on Lymphocyte Separation Media (in 1.5:1 ratio, MP Biomedicals 0850494X) and centrifuging at 500xg for 10 minutes and then collecting the buffy coat. The buffy coat cells were then pelleted and used for downstream analysis. Transfection was performed using Lipofectamine 2000 (Invitrogen) as described in the manufacturer's protocol. Normal growth medium was replaced 12-16 hours after transfection. Custom gapmeRs targeted to SNHG12 (Qiagen) or a negative control (Qiagen) and siRNA targeting IMP3(Origene, SR307255), YBX1 (Origene, SR303243), DHX9 (Origene, SR301175), and DNA-PK (OriGene, SR321433) were used for transfection. For murine in vivo gapmeR injections, 37.5 nmol naked gapmeR targeted to Snhg12 or control (Qiagen, LG00166878-DFA and LG00000002) was injected in 100 μl volume by tail vein injection with two-day loading protocol followed by biweekly injections until sacrifice. For intramuscular gastrocnemius gapmeR delivery in db/db mice, 37.5 nmol gapmeR was mixed with 25 μl Lipofectamine 2000 in 100 μl OptiMEM according to manufacturer's protocol and injected at three points (distal, mid and proximal gastrocnemius muscle).

Skeletal muscle histology
Skeletal myofiber cross sectional area (CSA) was assessed by immunofluorescence microscopy. Gastrocnemius muscles were sectioned, deparaffinized and rehydrated. Antigen retrieval was performed using an antigen unmasking solution (Vector Laboratories, H-3300). Muscle sections were then permeabilized with 0.25% Triton X-100 (ThermoScientific, 28313) for 10 minutes followed by three, two-minute washes in PBS. Sections were then blocked in PBS supplemented with 1% BSA and 5% goat serum for 1 hour at room temperature. The slides were then briefly washed with PBS and incubated with 5µg/µl wheat germ agglutin (ThermoScientific, W32466) for 30 minutes to label myofiber membranes. Coverslips were mounted with Vectashield hardmount containing DAPI (Vector Laboratories, H-1500). Images were obtained at 20x magnification using an Evos FL2 Auto microscope (ThermoScientific) and tiled images of the entire muscle section were used for analysis. Non-myofiber area was quantified by thresholding tiled images to obtain the pixel area of tissue between myofibers stained by wheat germ agglutin. Myofiber CSA were analyzed using MuscleJ (69), an automated image analysis software developed within Fiji.

Lentivirus production and transduction
Lentivirus for pUltra (Malcolm Moore, Addgene, 24129) was generated by co-transfection of 293T cells (ATCC) using Lipofectamine 3000 (Life Technologies) with pMD2.G (Didier Trono, Addgene) and psPAX2 (Didier Trono, Addgene) in a ratio 3:2:1, respectively. Transfection mix was added dropwise to dish and medium was changed 16hrs later. The supernatant was collected two days later by filtering through 0.45μm filter and stored at -80°C in 5 ml aliquots. Transduction of HUVECs was carried out in 6-well, 12-well, or 24 well plates by adding 1:1 lentiviral supernatant/medium in combination with 8 μg/mL polybrene (American Bio). Normal growth medium was replaced after 36 hours.

LncRNA pulldown
Biotinylated RNA was generated using T7 RNA polymerase kit (Thermo Scientific) by adding 1 μg linearized plasmid DNA, 10X biotin RNA labeling mix (Roche, 11685597910), 5x transcription buffer (Agilent) and RNasefree water in total volume of 20 μl and incubated for 2 hours at 37°C. Subsequently, DNase I (NEB) was added and the reaction was further incubated 15 min at 37°C to remove DNA template. The reaction was terminated using 0.8 μl 0.5M EDTA (pH 8.0). Purified biotinylated RNA was obtained using G-50 Sephadex Columns (Sigma Aldrich, 112739965001) according to the manufacturer's protocol. After RNA concentration was determined on NanoDrop (Thermo Fisher), the purified RNA was immediately used, or stored at -80°C. When needed, 10 pmol of biotinylated RNA (calculated using https://www.promega.com/resources/tools/biomath/) was heated for 2 min at 90°C in RNA structure buffer (10 mM Tris pH 7.0, 0.1 M KCl, 10 mM MgCl2). The mix was immediately transferred to ice and incubated 2 min and subsequently incubated at room temperature for 20 min.
For in vivo pulldowns of Snhg12-interacting proteins, biotinylated RNA was injected on consecutive days 1 and 2 by tail vein (15 μg/injection) before aortas were isolated on day 3. The tissue was processed as described above for cell lysate and nuclei were isolated using a Nuclear Extraction Kit (Millipore).

RNA synthesis, modification and injection
10 μg linearized and purified T7 vector with the cassette for LacZ or SNHG12 was used for 1x T7 RNA polymerase transcription reaction (Promega, RiboMax Large Scale RNA, PRP1300) based on the manufacturer's protocol. After 4 hours at 37°C, RNA was purified with phenol:chloroform isolation and resuspended in 140 μl RNase-free water.
After 5 minutes at 65°C, RNA was capped and 2'-O-methylated using Vaccinia Capping System (Fisher 50591120 and NEB M2080S) based on manufacturer's protocol before purification on Qiagen Universal Midi Kit RNeasy columns. RNA was stored at -80°C prior to use.

Clinical RNA-Seq Analysis
Using the published database by Ryan et. al. (accession number GSE120642) obtained from the GEO database through PubMed, the normalized counts file was used to generate relative expression dot-plot graphs with mean and standard error for various genes of interest that are involved in the angiogenesis response. Comparisons were made between healthy adult (HA), ischemic claudicant (IC) and critical limb ischemia (CLI) groups, by normalizing to the HA (at 100%) as a control cohort. Statistical comparisons were made using 1-way ANOVA. Data are displayed as relative expression.

RNA Isolation and RT-qPCR
Tissues (liver, gastroc) were homogenized using 5mm Stainless steel beads (Qiagen) TissueLyser II (Qiagen) according to manufacturer's instruction. For RNA isolation, TRIzol reagent (Invitrogen) or RNeasy kit (Qiagen) was used based on manufacturer's protocol. Subsequent RT-qPCR was performed using High-Capacity cDNA Reverse Transcription kit (Applied Biosystems). For SyberGreen based assay GoTaq qPCR Master Mix (Promega) was used.
Expression of mRNAs and lncRNA expression levels were normalized to HPRT, or β-actin (Aglient, AriaMx Real Time PCR System). Changes in expression were calculated using delta-delta Ct method.

RNA immunoprecipitation (RIP)
Nuclear pellets were prepared, harvested and homogenized by resuspending a pellet of 10 7 HUVECs in nuclear isolation buffer as noted above in LncRNA pulldown methods section. After homogenization, the supernatant was cleared by applying 40 μg magnetic protein A/G beads (Thermo Scientific) with non-specific rabbit IgG antibody (Invitrogen). The supernatant was collected and 50μL lysate was saved as an input control. The remaining lysate was divided into multiple portions of equal volume and 1 μg of protein-specific antibody was added and samples were incubated for 1 hour at 4°C in the presence of fresh magnetic protein A/G beads. Protein-specific antibodies used included IMP3 (Cell Signaling 57154S), YBX1 (Cell Signaling 4202S), DHX9 (Abcam ab26271), and DNA-PKcs (Abcam, 70250) or no antibody. Immuno-complexes captured by magnetic beds were washed three times with ice-cold RIP buffer and resuspended in Trizol (Invitrogen). RT-PCR was subsequently performed using the same input volumes.

RNA-Seq analysis
RNA-Seq analysis was performed after ribodepletion and standard library construction using Illumina HiSeq2500 V4 2x100 PE (Genewiz). All samples were processed using an RNA-seq pipeline implemented in the bcbio-nextgen project (https://bcbionextgen.readthedocs.org/en/latest/). Raw reads were examined for quality issues using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to ensure library generation and sequencing were suitable for further analysis. Trimmed reads were aligned to UCSC build mm10 of the mouse genome and augmented with transcript information from Ensembl releases 86 (H. sapiens) and 79 (M. musculus) using STAR (60). Alignments were checked for evenness of coverage, rRNA content, genomic alignment context and other quality checks using a combination of FastQC and Qualimap (61). Counts of reads aligning to known genes were generated by featureCounts (62). Differential expression at the gene level was called with DESeq2 (63). Total gene hit counts and CPM values were calculated for each gene and downstream differential expression analysis between specified groups was performed using DESeq2 and an adapted DESeq2 algorithm that excludes overlapping reads.
Genes with adjusted FDR< 0.05 and log2fold-change (>1.5) were called as differentially expressed genes for each comparison. The mean quality score of all samples was 35.91 with a range of 28,000,000-63,000,000 reads per sample. All samples had at least >96% of mapped fragments over total fragments. Coverage was visualized using Integrative Genomics Viewer (IGV) (version 2.3.68). Ingenuity Pathway Analysis/IPA (QIAGEN) and MetaCore (v20.2) were used for functional/gene set enrichment analysis. RNA-seq data will be available through the Gene Expression Omnibus upon publication. Figure 1S: In vitro angiogenesis assays using SNHG12 gain-and loss-of-function models. Figure 1S: SNHG12 expression in gain-and loss-of-function models by qRT-PCR. Cells treated as those in Figure 2 were isolated at 48 hours after transfection or transduction with either control or SNHG12 gapmeR or lentivirus. SNHG12 expression relative to HPRT RNA was measured by qRT-PCR by the method of delta-delta Ct. (n=4 per condition) ***P<0.001 using Student's t-test.  Figure 3S: Muscle fiber staining shows a leftward shift to smaller myofiber size in SNHG12 KD mice. A. Microscopy images of gastrocnemius muscle labeled with wheat germ agglutin (WGA) and DAPI show an increase in the percentage of small myofibers less than 500μm 2 or 1000μm 2 and expansion of non-myofiber area, features demonstrating increased ischemic myopathy in SNHG12 KD-treated C57BL6 mice. Scale bars = 200μm. *P<0.05, **P<0.01, ***P<0.001 using two-tailed t-test. * P<0.05 using Student's t test. B. Microscopy images of gastrocnemius muscle labeled with wheat germ agglutin (WGA) and DAPI show an increase in the percentage of small myofibers less than 500μm 2 and non-significant trend in expansion of non-myofiber area, features demonstrating increased ischemic myopathy in SNHG12 KD-treated db-db mice. Scale bars = 200μm. *P<0.05 using two-tailed t-test. Figure 4S: Tissue SNHG12 Expression Prior to Femoral Artery Ligation. Figure 4S: Tissue Snhg12 Expression Prior to Femoral Artery Ligation. C57Bl/6 or db/db mice were injected with gapmeRs on day 1 and day 2 and sacrificed on day 3. Liver, PBMCs, and gastrocnemius EC and non-EC fractions were isolated and RNA was purified for qRT-PCR analysis of Snhg12 compared to Hprt, which is normalized for each tissue. There is significant decrease in the expression of Snhg12 compared to Hprt in the EC and non-EC fractions of gastrocnemius muscle in db/db and non-EC of C57Bl/6 and a trend in EC from C57Bl/6. * P<0.05 using Student's t test.   Microscopy of fixed and sectioned gastrocnemius muscle from mice sacrificed on day 13 after systemic injection of FITC-lectin. CD31 and isolectin co-staining of vessels occurs in overlapping fashion in gastrocnemius. Control versus SNHG12 gapmeR-injected mice revealed a 23-28% reduction in average arterial diameter (by CD31 or isolectin, respectively) and a 46% reduction in the number of arteries per high-powered field (hpf) but no overall change in total CD31 or isolectin staining (5 sections per gastrocnemius, n=5 mice per group, scale bar = 100 μm). * P<0.05, ** P<0.01 using Student's t-test.